báo cáo khoa học: " Analysis of anther transcriptomes to identify genes contributing to meiosis and male gametophyte development in rice" pptx

RESEARCH ARTICLE Open Access
Analysis of anther transcriptomes to identify
genes contributing to meiosis and male
gametophyte development in rice
Priyanka Deveshwar
1
, William D Bovill
2
, Rita Sharma
3
, Jason A Able
2
and Sanjay Kapoor
1*
Abstract
Background: In flowering plants, the anther is the site of male gametophyte development. Two major ev ents in
the development of the male germline are meiosis and the asymmetric division in the male gametophyte that
gives rise to the vegetative and generative cells, and the following mitotic division in the generative cell that
produces two sperm cells. Anther transcriptomes have been analyzed in many plant species at progressive stages
of development by using microarray and sequence-by synthesis-technologies to identify genes that regulate anther
development. Here we report a comprehensive analysis of rice anther transcriptomes at four distinct stages,
focusing on identifying regulatory components that contribute to male meiosis and germline development.
Further, these transcriptomes have been compared with the transcriptomes of 10 stages of rice vegetative and
seed development to identify genes that express specifically during anther development.
Results: Transcriptome profiling of four stages of anther development in rice including pre-meiotic (PMA), meiotic
(MA), anthers at single-celled (SCP) and tri-nucleate pollen (TPA) revealed about 22,000 genes expressing in at least
one of the anther developmental stages, with the highest number in MA (18,090) and the lowest (15,465) in TPA.
Comparison of these transcriptome profiles to an in-house generated microarray-based transcriptomics database
comprising of 10 stages/tissues of vegetative as well as reproductive development in rice resulted in the
identification of 1,000 genes specifically expressed in anther stages. From this sub-set, 453 genes were specific to
TPA, while 78 and 184 genes were expressed specifically in MA and SCP, respectively. The expression pattern of

selected genes has been validated using real time PCR and in situ hybridizations. Gene ontology and pathway
analysis of stage-specific genes revealed that those encoding transcription factors and compon ents of protein
folding, sorting and degradation pathway genes dominated in MA, whereas in TPA, those coding for cell structure
and signal transduction components were in abundance. Interestingly, about 50% of the genes with anther-specific
expression have not been annotated so far.
Conclusions: Not only have we provided the transcriptome constituents of four landmark stages of anther
development in rice but we have also identified genes that express exclusively in these stages. It is likely that
many of these candidates may therefore contribute to specific aspects of anther and/or male gametophyte
development in rice. In addition, the gene sets that have been produced will assist the plant reproductive
community in building a deeper understanding of underlying regulatory networks and in selecting gene
candidates for functional validation.
* Correspondence:
1
Interdisciplinary Centre for Plant Genomics and Department of Plant
Molecular Biology, University of Delhi, South Campus, New Delhi - 110021,
India
Full list of author information is available at the end of the article
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>© 2011 Deveshwar et al; licensee BioMed Centr al Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://cre ativecommons.org/licenses/by/2.0), which permits unrestricted use, distribut ion, and
reproduction in any medium, provided the original work is properly cited.
Background
The anther is the male reproductive organ in flowering
plants and is composed of both reproductive and non-
reproductive tissues. The reproductivetissueoriginates
as a mass of primary sporogenous ce lls which are pro-
duced from the division of archesporial cells in the L2
layer of anther primordia. These cells divide mitotically
to give rise to the microspore mother cells (or meio-
cytes), that undergo m eiosis to produce haploid tetrads

of microspores [1]. This reductional division assures
genetic diversity in sexual reproduction via pairing and
recombination between homologous chromosomes.
Cytological ly, there are more commonalities than differ-
ences between the processes of mitosis and meiosis, e.g.,
condensation of chromosomes, their distinctive align-
ment at metaphase, followed by separation of sister
chromatids/homologous chromosomes at anaphase,
grouping of two nucleoids at telophase, and finally cyto-
kinesis that physically partitions the nucleo-cytoplasmic
compartments. Besides these similarities, there are a few
vital dissimilarities that distinguish these two processes,
including pairing and recombination of homologous
chromosomes during meiosis (which underlines the
basis of genetic diversity). This is followed by segrega-
tion of homologues and non-sister chromatids by unipo-
lar attachment of sister kinetochores to spindles, during
the first meiotic division. In the last decade, a number
of cell division components involved in chromosome
condensation, sister chromatid/homologous chromo-
some cohesion, kinetochore-spindle attachment/align-
ment, and cytokinesis have been identified. However, we
still know very little about the regulatory networks that
control the functioning of such components in a mito-
sis- or meiosis-specific manner.
Unlike in animals, haploid sperm are not produced
directly after meiosis in plants. Instead, the haploid
microspores are freed from the tetrad by the action of
callase, and then divide mitotically twice to produce a
three-celled functional male gametophyte known as pol-

len. The first mitosis is asymmetric which results in two
cells of different sizes and with dissimilar fates. The lar-
ger vegetative cell occupies most of the pollen space and
does not divide further but later, at the time of germina-
tion, forms the pollen tube. The smaller generative cell
undergoes o ne more round of mitotic division (symme-
trical this time) to produce two sperm cells. One of the
sperm cells fertilizes the egg cell in the female gameto-
phyte to form the zygote and the other fuses with the
two polar nuclei to form the triploid endosperm. Devel-
opment and release of mature pollen is dependent on
the elaborate coordination of m any genes expressed in
both non-reproductive as well as reproductive cell layers
of the anther. Thus, the anther is a multicellular organ
that undergoes complex processes such as cell fate
determination [2], cell differentiation, reductional divi-
sion [3] and cell-cell communication [4].
Our understanding of the genes that regulate de velop-
mental aspects of the anther is largely based on infor-
mation gathered by gene function knockdown
approaches, either by mutagenesis or RNA interference
(RNAi). Most of the pioneering research has been done
in Arabidopsis but at the same time many genes have
also been identified and characterized in rice revealing
gene function deviations or novel gene functions (for
reviews, see [5,6]). For example, the c haracterization of
an Arabidopsis EXCESS MICROSPOROCYTES 1 (EXS/
EMS1) ortholog ue in rice (MULTIPLE SPROPORO-
CYTES I - MSPI) and subsequent delineation of its
interaction with the TAPETUM DETERMINANT 1

(TPD1) rice orthologue (OsTDL1A), revealed its novel
function in restricting the number of sporogenous cells
in the ovule as well as in the anthers [2,7-10].
Although the gene knockout /knockdown approach (in
combination with the over/ectopic-expression approach)
can enable classification of a particular gene in context
of a biological phenomenon, these methods do not pro-
vide detailed information about the other components
of the regulatory circuitry that are positioned either
upstream or downstream in the hierarchy. Building a
regulatory network around this nucleation point is often
a difficult task that involves a combination of protein-
protein, DNA-protein and mutant analysis strategies.
However, analysis of transcriptome level perturbations
in developmentally or physiologically distinct states may
help in the segregation of var ious molecular contribu-
tors into co-expression groups, which could be further
analyzed for specific interactions [11,12]. Microarray-
based studies carried out in Arabidopsis [13], whe at [14]
and rice [15] have revealed the complexity of gene
expression during stages of anther development by use
of high density microarrays. Honys and T well [13] car-
ried out transcriptome analysis of male gametophyte
developm ent in Arabidopsis wheretheyidentifiedand
categorized microspore-expressed genes on the basis of
co-expression profiles. Of particular note is the study
conducted by Crismani and co-workers [14], w here
these authors used wheat Affymetrix GeneChip to moni-
tor the expression dynamics across seven stages of
anther development in the complex polyploid, bread

wheat. More recently, in rice, distinguishable differences
between the tapetum and male gametophyte transcrip-
tomes have been ascertained by using laser micro-
dissected cells of specific tissue types [16,17]. Collectively,
all these studies highlight the contrast of expression
between gametophytic and sporophytic tissues. How-
ever, because of the lack of comparison with other
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 2 of 20
tissue/cell-types most of these studies fall short of iden-
tifying genes tha t express specifically in these cell types
and, therefore, would almost certainly be playing signifi-
cant regulatory roles in controlling various aspects that
are unique to male gametophyte development.
An objective of the current study was to identify genes
that exhibit anther stage-specific expression patterns. To
achieve this we performed whole genome microarray
analysis on rice anthers isolated at pre-meiotic (PMA),
meiotic (MA), single-celled microspore ( SCP), and tri-
nucleate pollen (TPA) stages of development. Since
whole anthers were used in this study, we expected the
data to include contributions from all cell t ypes. We
performed differential expression analysis to identify
genes regulating precise developmental events during
anther development. By including transcriptomic data of
four vegetative and seed developmental stages/tissue
types in the differential expression analysis, we have
attempted to identify and segregate expression profiles
specifically (preferentially) relevant to the events related
to male gametophyte development. These analyses have

identified genes that express specifically in P MA, MA,
SCPandTPA.Furthermore,thedatahavealsobeen
analyzed for the expression specificities of known meio-
sis-related genes and those contributing to sperm cell
transcriptomes in other systems. Our data therefore pro-
vides a firm founda tion for future investigations cen-
tered on delineating the molecular networks of male
meiosis, early gametophyte development and sperm cell
differentiation in rice.
Methods
Tissue collection and RNA extraction
Wild type rice (Oryza sativa subsp. indica cv. IR64) was
transplanted in fields in mid-June, 2007. Temperature
ranged from 35-40°C
max
and 25-29°C
min
.Aconstant
wat er supply was available throughout the growing per-
iod. Tissue was harvested at different stages of anther
development from about 30 to 60 days after transplant.
Florets at various stages of development were dissected
using a Leica MZ 12.5 (Leica Gmbh, Wetzlar, Germany)
dissecting microscope to collect anthers. Anther
squashes were prepared from one representative anther
in each floret, stained with DAPI, and observed under a
fluorescence microscope (DM 5000B, Leica Gmbh, Wet-
zlar, Germany) to confirm the developmental stage
according to Raghvan [18]. Anthers isolated from 8-10
plants were bulked into three biological replicates.

After collection and staging into separate groups con-
taining four developmentally distinct stages [pre-meiotic
anther (PMA; from the first i dentifiable anther like
structure to the end of interface), meiotic anther (MA;
leptotene to tetrad), anthers with single celled pollen
(SCP) and anthers with tri-nucleate pollen (TPA); Table 1],
anthers were plac ed in Trizol reagent (Invitrogen, CA,
USA) and kept at -70°C until RNA isolation. High quality
RNA, assessed by a bio-analyzer (Agilent, CA, USA), was
used for hybridization experiments with the 57K Rice
Genome Array (Affymetrix, CA, USA).
Microarray experiments
A total of 3 μgoftotalRNAisolatedfromantherswas
amplified and labeled using a one-cycle target labeling
kit (Affymetrix, CA, USA). Target preparation, hybridi-
zation, washing, staining and scannin g of the chips were
done according to the manufacturer’ sprotocol.Gene-
Chip
®
Operating Software 1.2.1 (GCOS) was used for
washing and staining of the chips in a Fluidics Station
450 (Affy metr ix, CA, USA) and scanned with a Scann er
3300 (Affymetrix, CA, USA). Three biological replicates
processed for each stage with overall correlation co-
efficient values of more than 0.97 were further used for
the final data analysis, which underlines the high repro-
ducibility and reliability of the microarray data.
Microarray data analysis
CEL files for four anther development stages generated
by GCOS were transferred to ArrayAssist ver. 5.5 (Stra-

tagene, CA, USA) microarray data analysis software for
analyses. A combined project was made where CEL files
of the four anther s tages, as well as those for mature
leaf, Y-leaf, root, 7-day-old seedling, shoot apical meris-
tem (SAM; meristematic tissue isolated from the apex of
the shoot from plants in which more than half of the til-
lers already had panicles) and five stages of seed devel-
opment (S1, S2, S3, S4 and S5), have been deposited to
the Gene Expression Omnibus (GEO; i.
nlm.nih.gov/geo/; accession numbers GSE6893 and
GSE6901).
Table 1 Classification of rice panicles and florets for
categorization of anther development stages
Anther Development
(PMA)
Pre-
meiotic
anther
(MA)
Meiotic
anther
(SCP) Anther
with single
celled pollen
(TPA) Anther
with tri-
nucleate
pollen
Anther
development

stage [47]
Stage 3-5 Stage 6-8 Stage 9-10 Stage 12-14
Anther length
(mm)
0.35-0.45 0.50-0.85 0.90-0.95 2.0-2.5
Floret length
(mm)
1.5-2.5 3.5-6.0 7.0-7.5 >8.0
Panicle
length (cm)
1.0-5.0 6.0-11.0 8.0-15.0 25.0-30.0
Note: Panicle, floret and anther length indexing is standardized only for IR64
cultivar of Oryza sativa subsp. indica, and may vary in different cultivars of rice.
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 3 of 20
The rice Affymetrix GeneChip
®
contains 57,381
probe-sets, however, not all of the probe-sets corre-
spond to annotated genes, and in some instances more
than one probe-set corresponds to annotated genes.
Therefore, in order to identify the unique probe-sets
that correspond to annotated genes, the MSU Rice
Pseudomolecule ( />data/Eukaryotic_Projects/o_sativa/annotation_dbs/) ver-
sion 5, KOME ( and
NCBI ( databases were
used, with the probe-set list manually curated. Conse-
quently, a total of 3 7,927 probe-sets were identified a s
unique non-redundant probe-set IDs (after removing
hybridization controls, transposable element (TE)

related g enes, redundant probe-sets and probe-sets
without a corresponding locus in the databases men-
tioned above). All subsequent expression analysis was
carried out on this reduced dataset. The MAS5 algo-
rithm was applied (with default parameters) to identify
genes that could be classified as expressed or non-
expressed. 66% present calls in a triplicate (as PPP,
PPA or PMM) dataset were kept as minimum criteria
for a gene being ‘ expressed’ or otherwise ‘ non-
expressed’. The microarray data was normalized using
the GC-RMA algorithm followed by log
2
transforma-
tion. To identify differentially expressed genes, one-
way Analysis of Variance (ANOVA) was performed on
the four anther development stages with the Benjamini
Hochberg correction [19]. Further, a stringent statisti-
cal criterion of a t least a 2-fold change at a p-value
≤0.005 was used for gene selection. Cluster analy sis
was performed using the K-means clustering algorithm
ofArrayAssist(Stratagene,LaJolla,CA,USA).Allthe
heat-maps were made using GC-RMA log transformed
sample averages.
Expressio n values of probe-sets of Magnoporthe genes
present on the chip were used as a criterion to define
“ absent” genes (Additional File 1) since their signal
value should represent the background signal. Average
of the median for those genes plus 5 i.e., 10 GC-RMA
value was put as the upper limit for a gene to be called
‘absent’. Annotations for functional a lignment of genes

were retrieved from Osa1 Rice Genome Annotation Pro-
ject release 6 (RGAP: />Identification of putative orthologues in rice
We have previousl y described the identifica tion of puta-
tive rice orthologues of meiotic genes [20]. Briefly, the
sequences of Saccharomyces cerevisiae and Arabidopsis
thaliana genes involved in double strand break (DSB)
formation, recombination, synaptonemal complex
assembly, chromosome pairing and DNA mismatch
repair were used as queriesforTBLASTXanalysis
against all green plants at The Institute for Genomic
Research’s (TIGR) Plant Transcript Assembly (TA) data-
base. A significance value of >E
-20
from the TBLASTX
analysis was used to identify putative orthologues in
wheat, rice and barley. The rice TA IDs for meiotic
gene orthologues [20] were used to identify the corre-
sponding rice Osa1 loci (MSU Rice Genome Annotation
(Osa1) Release 6.1; ) and
their respective Affymetrix probe-sets, which were used
for e xpression analysis. For the identification of sperm-
expressed genes, cDNA and EST sequences of Arabi-
dopsis,maizeandlilyweredownloadedfromTAIR
( and NCBI (http://www.
ncbi.nlm.nih.gov/). These sequences were used as
queries for BLASTx against a local database made with
the Osa1 Release 6.1 Rice proteins using BIOEDIT soft-
ware ( />with a signif icance value of > E
-20
used for identifying

rice orthologues (Additional File 2).
Real-time quantitative PCR (Q-PCR)
cDNA for the real-time reactions were synthesized using
the same RNA samples that were used for microarrays.
Real-time PCR primer designing, reactions and calcula-
tions were carried out as described previously [21]. Pri-
mers used in the experiment are listed in Additional
File 3.
In situ hybridizations
Florets were fixed in FAA (10% formaldehyde, 5% acetic
acid and 50% ethanol) for 24 hours at 4°C and then
dehydrat ed in a graded ethanol series followed by a ter-
tiary butanol series, before placing in paraplast plus
(Sigma Aldrich). Paraplast embedded florets were sec-
tioned by using a Leica RM2245 rotary microtome pro-
ducing 8 μm thick sections that were placed on Poly-L-
Lysine coated slides (Polysciences Inc.). Approximately
200 bp sequences from the genes LOC_Os04g52550 and
LOC_Os01g70440, were a mplified using primers (for-
ward 5’ -CAT GTTCTTCCTCTGACGACA-3’ and
reverse 5’ -GACACGGACAAAAATTTACTATGG-3’ )
and (forward 5’ -CTCCACCTCGC TCTG ATTAA- 3’ and
reverse 5’-TCATTTCAATGCAGTACAGGC-3’), respec-
tively. These clone d products were then ligated into the
pGEMT-Easy vector (Promega). The clones were linear-
ized with Sal I and Nco I enzymes for in vitro transcrip-
tion of digoxinin labeled RNA probes with T7 and SP6
RNA polymerase, respe ctively, according to the manu-
facturer’s instructions (Roche). The in situ pre-treatment
and hybridization steps were essentially carried out as

described [22]. Immunological detection was carried out
using the Roche DIG detection kit, followi ng the manu-
facturer’ s protocol. Sections were mounted in DPX
mounting medium and observed under the microscope
(DM 5000B, Leica Gmbh, Wetzlar, Germany).
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 4 of 20
Results
Development-dependent changes in the anther
transcriptome
Transcriptome profiling of anther development required
isolation of anthers at landmark stages of development,
i.e., pre-meiosis (PMA), meiosis (MA), immediately after
meiosis where single-celled microspores are relea sed
from tetrads (SCP) and mature anthers with tri-nucleate
pollen (TPA) just prior to dehiscence. For this, the rice
florets were initially broadly classified on the basis of
their size and then one anther from each floret was
microscopically examined to confirm the stage of male
gametophyte development by staining with DAPI before
staging the rest into one of the four classes specified
above(Table1).Microarraydatafromthethreerepli-
cates of each stage exhibited correlation co-efficients of
0.99 (PMA), 0.99 (MA), 0.99 (SCP) and 0.97 (TPA).
Scatter plot analysis was performed to analyze the extent
of transcriptome level variations between the four
anther stages (Additional File 4). Interestingly, PMA,
MA and SCP showed high correlation values between
0.92-0.96, however, TPA was found to be markedly dif-
ferent in its transcript constitution from the other stages

of anther development, with correlation co-efficients
ranging between 0.77 and 0.79. This difference was also
reflected in the number of differentially (2-fold at p-
value ≤ 0.005) regulated genes (7219-8318 between TPA
and other anther stages). To determine the extent of
transcriptome level changes that are required for anthers
to differentiate from the undifferentiated meristematic
cells, the PMA transcriptome was compared with that
of the shoot apical meristem (SAM). The SAM and
PMA showed significant correlation (0.94), which gradu-
ally declined with the progression of anther develop-
ment to 0.90 (SAM:MA), 0.87 (SAM:SCP) and 0.73
(SAM:TPA).
The oligonucleotide probes on the rice Affymetrix
Genome Array represent 37,927 unique genes including
33,813 gene loci mapped in MSU Rice Genome Annota-
tion Release 6 and 4,114 unique, but unmapped, cDNA/
ESTs (KOME and NCBI). This represents 93.5% of the
latest estimates of 40,577 non-TE-related protein-coding
genes on the rice pseudomolecules. To define the extent
of the anther transcriptome, t he expressed genes were
differentiated from the non-expressed genes (see Materi-
als and Methods). Consequently, 21,597 genes were
identified as expressed in at least one stage of anther
development (Figure 1a). MAS5 detection calls and their
p-values are given in Additional File 5. The MA and
SCP stages were found to express the maximum number
of genes, i.e., 18,090 and 17,953, respectively. Number of
genes specifically present amongst anthers was identified
as those where expression in all the other anther sta ges

except o ne had GC-RMA expression value less than 10
( see Materials and Methods). The TPA transcriptome
was the smallest with 15,465 expressed genes but it
represented the most diverse transcriptome with the lar-
gest pro portion (4.4%) of genes expressed specifically at
this developmental stage amongst anthers. The propor-
tion of specifically expressed genes was found to be
2.0%, 0.5% and 0.3% in SCP, MA and PMA, respectively.
The cumulative anther transcriptome was compared
with the previously generated transcriptomes of root,
leaf and five stages of seed development of the same
rice cultivar [21,23] to identify the extent of overlap
between various transcriptomes (Figure 1b). In total,
Anther
(21,597)
Anther and SAM
(22,115)
Anther
(21,597)
Seed
(21,062)
Leaf
(16,416)
Root
(18,166)
)
14,121
2,295
707
246

1,246
62
155
396
419
353
369
1,034
1,504
762,016
(a)
(b)
Number of genes
Percent specific amongst anthers
8000
9000
10000
11000
12000
13000
14000
15000
16000
17000
18000
19000
SAM PMA MA SCP TPA
0.0
1.0
2.0

3.0
4.0
5.0
6.0
7.0
8.0
9.0
10.0
16719
17497
18090
17953
15465
4.4%
2.0%
0.5%
0.3%
0.7%
Figure 1 Transcriptome profile of anther development.
(a) Anther development transcript sizes overlaid with a line graph
depicting the percentage of specifically expressed genes in
individual stages. The figure highlights that the meiotic anthers
have the largest transcriptome, whereas, anthers at the tri-nucleate
stage of pollen development show a comparatively smaller
transcriptome, but with the largest proportion of specific genes.
(b) Venn diagrams showing the constitution of vegetative tissues
(leaf and root), seed and anther transcriptomes with component
overlaps amongst them.
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 5 of 20

14,121 genes express in all the stages analyzed, suggest-
ing their involvement in housekeeping functions or gen-
eral metabolism. This analysis also highlighted that
anthers have the largest (21,597 genes) and the most
diverse transcriptome of all the stages analyzed, as
expression of 2,295 (10.6%) genes was unique to anthers.
In comparison, the numbers of uniquely expressed
genes in roots, leaves and seeds were 707, 246 and
1,246, respectively. Besides identifying 14,121 commonly
expressed genes between all four developmental stages,
the anther transcriptome shared maximum similarity to
that of the seed transcriptome with 4,554 commonly
expressed genes in anther and seed stages. However, a
much lower level of similarity between the anther and
root (2,488), and anther and leaf (1,265) transcriptomes
was observed.
Co-regulated clusters of differentially expressed genes
To identify genes with similar expression profiles during
anther development, the normalized expression data was
subjected to one-way ANOVA that resulted in the selec-
tion of 14,672 differentially expressed genes at a p-value
≤0.005. Using a cut- off of 2-fold change in expression in
any stage of anther development further filtered these
genes to 11,915 (Additional File 6). Using K-means clus-
tering, these genes could be clustered into 10 major
groups, which were further categorized into sub-groups
depending on the amplitude of expression (Figure 2).
Clusters 2 to 5 consisted of 8,014 (67.3%) differentially
expressed genes expressing in all stages of anther devel-
opment. Of these, only one gene was found to be speci-

fic to anther stages. Genes in these clusters either
showed up (cluster 4 and 5) or down regulation (clus-
ters 2 and 3) in TPA, while in other stages the differ-
ence in expression of these genes is not as significant. In
contrast, t he 733 (6.2%) genes in cluster 7 showed high
expression in PMA, MA and SCP; 571 (4.8%) genes in
cluster 9 were activated specifically in SCP, while clus-
ters 8 (372 genes; 3.1%) and 10 (1,071 genes; 9.0%)
exhibited MA- and TPA-prefer ential expression profiles,
respectively.
For the identification of specifically expressed genes dur-
ing anther development, five vegetative stages (mature
leaf,Yleaf,root,7dayoldseedlingandSAM)andfive
stages of seed development (S1, S2, S3, S4, S5) were
compared with anther stages. From the 11,915 differen-
tially expressed genes (from Figure 2), those with GC-
RMA normalized signal values less than or equal to 10
in vegetative and seed stages were filtered out (see Mate-
rials and Methods for criteria on ‘absent ’ genes). Genes
obtained w ere further filtered by identifying those with
at least a 2-fold higher signal value in any of the anther
stages than the highest value in the vegetative or seed
stages (i.e. these candidates would have at least a 20
GC-RMA signal value). After such strin gent filtering
1,000 anther-specific genes were identified (Figure 3).
Forty-five percent (45.3%) of them were on ly specifically
expressed in TPA, further emphasizing the distinctness
of this stage. SCP and MA have only 18.4% and 7.8% of
the specifically expressing genes respectively, while PMA
has a low share of stage specificity with 2.7% representa-

tion. Notably, those specifically expressed in PMA have
lower expression compared to other anther stages.
Percentages of anther specific genes were calculated
for each of the k-means clusters (Figure 2). Interestingly,
expression of 33.3% (914 genes) of the 2,747 genes in
clusters 7 to 1 0 was found to be specific to anthers. Of
these 914 genes, 138 (15.1% ) were specific to meiotic
anthers, 226 (24.7%) to anthers at the SCP stage, while
the largest group was expressed s pecifically at the TPA
stage (522 genes; 57.1%) (see Additional File 6).
Thedifferentiallyexpressedgenesineachofthe10
clusters were assigned to 19 functional categories and
those that could not be affiliated to any of these cate-
gories or that have not been annotated as yet were cate-
gorized as ‘ Others’ (approximately 34%; Ta ble 2).
Cluster-wise over representation of the number of genes
by 20% (taken arbitra rily as a measure of predominance)
of their overall percentage in individual functional cate-
gories has been highlighted to facilitate better visual
interpretation of the data (Table 2). Genes involved in
protein metabolism, involving fol ding, sorting and
degradation (6.9%), signal transdu ction (8.3%) and tran-
scription factors (7.1%) constitute three major functional
categories of differentially expressed genes during anther
development. Clusters 1, 2 and 3, which exhibited
down-regulatory trends from SAM to TPA (see Figure
2), were dominated generally by transcription factor,
chromatin remodeling, RNA metabolism, translation-
and cell cycl e-related genes. Expression profiles in clus-
ters 6b and 7, showing up-regulation in MA and SCP

followed by down-regulation in TPA, coincide with the
pattern of tapetum development. Coincidently, the
genes exhibiting these profiles were found to have over-
representation of those involved in carbohydrate, energy
and li pid metabolism, along with those involved in
transporter activities and vesicular trafficking. Cluster
10, which represents TPA specific expression profiles,
had an over-representation of genes involved in cell
structure, secondary metabolism, transporter activity
and signal transduction.
Validation of specific expression profiles by Q-PCR and in
situ hybridizations
To validate the microarray data, eight genes showing
specific expression in one or more stages of anther
development were selected for real-time/quantitative
PCR analysis (Figure 4). These include: one gene from
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 6 of 20
652
Cluster 1
SAM PMA MA SCP TPA
2
4
6
8
10
12
14
1.4
8

8.5
11
5.5
4
399669
SAM PMA MA SCP TPA
SAM PMA MA SCP TPA
SAM PMA MA SCP TPA
Cluster 5
(a)
(b)
2
4
6
8
10
12
14
2
4
6
8
10
12
14
1/0.3%
0/0.0%
311
191
SAM

2
4
6
8
10
12
14
PMA MA SCP TPA
SAM PMA MA SCP TPA
C
l
uste
r
6
(a)
(b)
2
4
6
8
10
12
14
SAM PMA MA SCP TPA
17/5.5%
2/1.0%
550 183
SAM PMA MA SCP TPA
SAM PMA MA SCP TPA
C

l
uste
r 7
(a)
(b)
2
4
6
8
10
12
14
2
4
6
8
10
12
14
23/12.6%
5/0.9%
781
602
SAM PMA MA SCP TPA
SAM PMA MA SCP TPA
Cluster 4
(a)
(b)
2
4

6
8
10
12
14
2
4
6
8
10
12
14
0/0.0%
0/0.0%
395
SAM PMA MA SCP TPA
C
l
uste
r
9
(a)
(b)
2
4
6
8
10
12
14

176
SAM PMA MA SCP TPA
2
4
6
8
10
12
14
50/28.4%
176/44.6%
497
1116
826
SAM PMA MA SCP TPA
SAM PMA MA SCP TPA
SAM PMA MA SCP TPA
Cluster 3
(a)
(b)
(c)
2
4
6
8
10
12
14
2
4

6
8
10
12
14
2
4
6
8
10
12
14
0/0.0%
0/0.0%
0/0.0%
372
SAM PMA MA SCP TPA
C
l
uste
r
8
(a)
2
4
6
8
10
12
14

138/37.1%
560
SAM PMA MA SCP TPA
C
l
uste
r 1
0
(a)
(b)
2
4
6
8
10
12
14
511
SAM PMA MA SCP TPA
2
4
6
8
10
12
14
291/56.9%
231/41.3%
66/10.1%
1231

1222
671
SAM PMA MA SCP TPA
SAM PMA MA SCP TPA
SAM PMA MA SCP TPA
Cluster 2
(a)
(b)
(c)
2
4
6
8
10
12
14
2
4
6
8
10
12
14
2
4
6
8
10
12
14

0/0.0%
0/0.0%
0/0.0%
Figure 2 Gene expression patterns of differentiall y expressed genes in SAM and the four stages of anthe r development (PMA, MA,
SCP, TPA) categorized into 20 groups using the K-means clustering tool. Groups with similar expression patterns but different expression
amplitudes have been grouped together to make 10 clusters. The normalized log transformed signal values were plotted for each of the five
stages. The number of genes in the clusters is indicated along the side of the heatmap. The percentage of anther-specific genes in each cluster
is specified at the lower left side of the heatmap.
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 7 of 20
cluster 3b exhibiting PMA specific expression; two genes
from cluster 7a and one gene from cluster 7b with high
and low expression, respectively, in MA and SCP; two
from cluster 8a with MA preferential expression; and
two genes from cluster 10a with expression mainly in
the TPA. Two of the selected gene s have been pre-
viou sly character ized and their reported expression pro-
files also matched with our analysis (OsMEL1 [24], RTS
[25]). Overall gene expression as identified by the micro-
array experiments, exhibited a high degree of similarity
with that obtained from the Q-PCR analyses with a
correlation co-efficient (r) greater than 0.9, thereby indi-
cating the reliability and robustness of the microarray
data.
Further, we validated our microarray e xpression
results by doing in situ hybridization of two of the
genes already validated by Q-PCR (Figure 5a). The tran-
scripts from LOC_Os04g52550, which codes for an
argonaute protein, were found to localize in the meio-
cytes as well as wall layers of meiotic anthers. Later in

development (SCP stage), the expression was found to
be restricted to the tapetum, microspores and vascular
ML
YL
Root
Sdl
SAM
PMA
MA
SCP
TPA
S1
S2
S3
S4
S5
1.69 15.428.56
Number of genes
27
27
35
3
23
10
4
12
78
49
12
23

184
60
453
1000
PMA TPASCPMA
(a) (b)
Figure 3 Expression profiles of specifically expressed genes in anthers. (a) Hierarchical cluster diagram representing expression patte rns of
1000 genes that show transcript accumulation in at least one of the four stages of anther development and undetectable expression in any of
the vegetative (ML, mature leaf; YL, Y-leaf; Root; SDL, 7-day-old seedling) or seed development stages (S1-S5; encompassing 0-30 days of seed
development after pollination). (b) A diagrammatic representation of the anther-specific expression profiles with the number of genes under
each expression profile.
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 8 of 20
tissue in the connective. LOC_Os01g70440, coding for a
LEM-1 family protein, exhibited expression in the tape-
tal laye r of anthers at tri-nu cleate stage with no expres-
sion in the pollen grains. The expression of both the
genes was restricted to anthers as no expression was
seen in lemma and palea (Figure 5a). We a lso scanned
the literature for in situ experiments where we could
correlate our ant her-specific or anther-preferential
expression with that reported previously. A s ummary of
expression domains of six such genes coding for OsC6
[26], OsMSP1 [9], OsRAD21-4 [27], OsMEL1 [24],
PAIR2 [28] and TDR [29] and their correlation with the
microarray expression profiles obtained from our dataset
is shown in Figure 5b. The in situ expression patterns of
two genes analyzed here and the six previously reported,
show good correlation with our microarray based pro-
files and subsequent differential expression analysis.

Developmental stage-wise activation/up-regulation of
genes
As anther developm ent progresses from PMA to TPA, a
number of processes are accomplished in a sequential
manner. By comparing gene expression between two
adjacent stages of anther development, we aimed to
identify the molecular components involved in switching
from one phase of development to the next. The res ults
of this comparative analysis where differ ences in expres-
sion between SAM:PMA, PMA:MA, MA:SCP, and SCP:
TPA stages wer e analyzed by setting the criteria of
2-fold change at a p-value ≤0.005 are shown in Figure 6a.
Only a small proportion of genes (624), were found to be
differentially ac tivated (319) or down-r egulated (305) in
PMA when compared to SAM. However the number of
differentially expressed genes steadily increased to 1,762
in MA, 3,376 in SCP and 7,251 in TPA in relation to
their respective previous stage of development. A greater
number of genes were up-regulated in comparison to
those down-regulated in PMA and MA, however, this
trend reversed in SCP and TPA where a larger propor-
tion of genes showed down-regulation (Figure 6a). This
finding might point towards a major post-meiotic switch-
ing of gene expression from the sporophytic to the game-
tophytic mode.
The stage-wise up-regulated genes during progression of
anther development were furtherminedforthosethat
were specifically activated in a particular stage (Figure 6a).
For this, specific genes with no detectable expression in
any previous anther stage were considered as specifically

Table 2 Association of differentially expressed genes in co-expression clusters (see Figure 2) with GO functional
categories
Percentage of transcripts classified in co-expression profiles in Figure 2.
Functional Categories
High to low Low to high PMA, MA, SCP, TPA PMA, MA, SCP MA SCP TPA
1 2 3 4 5 6 7 8 9 10 Total
Amino acid metabolism 0.6 1.6 1.1 1.4 1.8 0.8 1.1 0.3 0.2 0.6 1.2
Carbohydrate metabolism 0.8 2.0 1.2 4.4 1.4 3.4 2.5 1.1 1.4 1.6 2.0
Catalytic activity 1.7 2.7 2.4 4.8 3.8 6.8 5.7 4.0 3.0 3.9 3.5
Cell cycle 2.3 3.3 6.1 2.1 2.3 1.2 1.2 1.6 0.7 1.3 3.0
Cell structure 1.7 2.0 3.0 1.7 1.4 2.4 2.6 2.2 2.5 5.4 2.5
Chromatin remodeling 1.7 2.7 2.5 1.3 1.0 1.0 0.7 0.5 0.4 0.6 1.7
Energy metabolism 1.5 3.6 1.6 4.7 1.8 3.4 3.0 2.7 2.5 1.3 2.7
Lipid metabolism 1.1 2.2 2.5 3.6 3.0 6.6 4.4 3.0 4.6 2.6 2.9
Nucleotide metabolism 0.6 1.3 1.6 1.5 1.1 0.0 0.7 0.0 0.7 0.4 1.1
Protein-protein interaction 2.3 3.3 2.7 4.4 2.3 2.8 2.6 2.7 4.0 2.1 3.0
Protein metabolism 5.7 7.9 6.3 6.8 7.3 4.4 6.3 8.9 11.6 4.0
6.9
RNA metabolism 7.7 7.5 7.4 2.2 3.8 1.2 1.9 1.1 1.4 1.0 4.8
Secondary metabolism 1.8 0.6 0.9 0.9 1.2 3.8 4.0 3.2 5.3 2.7 1.7
Signal transduction 6.6 7.5 8.8 9.0 8.2 7.8 9.3 7.8 6.8 10.5
8.3
Stress 4.4 3.1 3.0 3.5 3.7 6.6 6.5 4.0 5.1 4.6 3.8
Transcription factors 8.9 8.0 7.7 5.3 5.1 7.0 7.6 8.9 6.1 5.7
7.1
Translation 1.1 8.8 2.8 1.5 1.3 0.2 1.0 0.0 0.7 0.2 3.4
Transporters 2.3 4.1 3.5 4.8 5.5 7.8 4.4 4.0 5.4 6.2 4.5
Vesicular trafficking 0.3 2.9 1.0 5.0 2.4 2.6 0.7 0.0 0.4 2.2 2.1
Others 46.9 25.1 33.9 31.1 41.3 30.5 34.0 44.1 37.5 43.1 33.8
Genes in each cluster 652 3124 2439 1383 1068 502 733 372 571 1071 11915

The total representation of genes (% values) of three major functional categories (besides ‘Others’) is shown in bold & underlined text. Over-representation of
genes in each functional category by more than 20% of their overall representation in individual clusters is highlighted with bold and italicized letters.
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 9 of 20
activated/triggered. Interestingly, only 33 genes (t hat is,
10.3% of 320 PMA up-regulated genes) were found to be
triggered in PMA. The percentage of specifically activated
genes ranged between 12 to 16% of the total up-regulated
genes in MA, SCP and TPA vis-à-vis their respective pre-
vious stage of development, with the number in the
respective stages being 133, 191 and 448. Functional asso-
ciation of stage-wise activated and 2 fold up-regulated
genes based on Gene Ontology (GO) annotations high-
lighted the molecular processes/components involved
(Figure 6b). Major perturbations in transcript abundance
were observed in genes coding for transcription factors,
signal transduction and cell structure components, cataly-
tic activity and those involved in the function of protein
folding, sorting and degradation. A significant number
(45) of genes coding for signal transduction components
were specifically activated in TPA, which may contribute
to the pollen-specific transcriptome involved in pollen-
pistil interactions and pollen tube growth. The largest
numbers of genes involved in protein metabolism were
triggered in the SCP stage, which coincided with the most
active phase of tapetal cells and their degeneration. Out of
the 88 cell structure related genes up regulated in TPA, 34
were specifically tr iggered at this sta ge that comprises
7.6% of the TPA triggered genes. This suggests most of
the up-regulated cytoskeletal genes may have a TPA speci-

fic function; most likely in pollen germination.
Expression dynamics of meiosis-related genes
The functional conservation of meiosis between eukar-
yotes can be exploited to identify new candidates for
meiotic regulation in rice. We have previously compiled
a database of yeast and Arabidopsis genes involved in
meiosis, and identified putative orthologues in the rice,
0
2
4
6
8
10
12
14
PMA MA SCP TPA
LOC_Os02g02820
r = 0.987 (gr-7a)
-2
0
2
4
6
8
10
12
PMA MA SCP TPA
LOC_Os09g16010
r = 0.985 (gr-8a)
0

2
4
6
8
10
PMA MA SCP TPA
LOC_Os10g24050
r = 0.970 (gr-7b)
0
2
4
6
8
10
12
PMA MA SCP TPA
LOC_Os04g52550
r = 0.986 (gr-8a)
0
2
4
6
8
10
12
14
PMA MA SCP TPA
LOC_Os03g58600 (OSMEL1)
r = 0.93 (gr-3b)
0

2
4
6
8
10
12
14
PMA MA SCP TPA
LOC_Os01g70440 (RTS)
r = 0.990 (gr-10a)
-2
0
2
4
6
8
10
12
14
PMA MA SCP TPA
LOC_Os12g23170
r = 0.961(gr-10a)
0
2
4
6
8
10
12
14

16
PMA MA SCP TPA
LOC_Os08g43240
r = 0.994 (gr-7a)
Microarray

QPCR
Figure 4 Q-PCR analysis of eight genes showing anther developmental stage-specific expression and its correlation with microarray
data. Three biological replicates were taken for both Q-PCR and microarray analysis. The Y axis represents normalized log
2
transformed
expression values obtained using microarray analysis and log
2
transformed relative transcript amount obtained by Q-PCR. The Q-PCR data has
been scaled such that the maximum expression value of Q-PCR equals that of the maximum value of the microarray to ease profile matching.
Gene locus IDs and their affiliation to the co-expression groups shown in Figure 3 are mentioned. The correlation co-efficient (r) between the
two expression profiles is also indicated. Expression of 18S rRNA was used as an internal control to normalize the Q-PCR data. PMA; pre-meiotic
anthers, MA; meiotic anthers, SCP; anthers with single-celled pollen, TPA; tri-nucleate pollen containing anthers.
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 10 of 20
(a)
(b)
LOC_Os04g52550
LOC_Os01g70440
Antisense
(MA)
Gene
Name
Locus and Affy IDs Reference
In-situ mRNA

localization
anther
)
Microarray
profile
OsC6
LOC_Os11g37280
Os.52076.1.S1_at

[26]
Strong expression in tapetal
cells and weak expression
in microspores of anthers at
stages 10 and 11
SCP

OsMSP1
LOC_Os01g68870
Os.23868.1.S1_at

[9]
Inner wall layer of anthers of
flower at stage II and PMCs
entering into meiotic
prophase

PMA-MA

OsRAD21-4
LOC_Os05g50410

Os.33045.1.S1_a_at

[27]
Highest in premeiotic PMCs
and relatively less in meiotic
PMCs and tapetal cells
PMA-MA

OsMEL1
LOC_Os03g58600
Os.40026.1.S1_at

[24]
Archesporial cells and
sporogenous cells of male
reproductive organs

PMA

Pair 2
LOC_Os09g32930
Os.49778.1.S1_at

[28]
Anther in meiosis

MA

TDR
LOC_Os02g02820

Os.50000.1.S1_at

[29]
Tapetal, middle layer, and
endothecium of the meiosis
stage anthers. At the tetrad
and young microspore stage,
more strongly expressed in
the tapetum
MA-SCP

PMA
MA
SCP
TPA
0
2
4
6
8
10
Antisense
(SCP)
Antisense
(TPA)
Sense
(TPA)
Sense
(MA)
PMA MA SCP TPA

0
2
4
6
8
10
12
PMA MA SCP TPA
T
T
V
MEI
W
M
Bar=100μm
L
L
L
M
Equivalent
stage(s
Figure 5 Validation of microarray data by in-situ hybridization.(a)In-situ localization of transcripts corresponding to t he genes
LOC_Os04g52550 and LOC_Os01g70440 in rice florets (MA, SCP and TPA stages as marked). Corresponding microarray-based expression profiles
of these two genes are also shown as bar graphs for comparison. W, wall layers; V, vascular tissue; T, tapetum; M, microspores; MEI, meiocytes; L,
lemma. (b) A compilation of in-situ localization analyses for six genes using published literature and their correlation with anther preferential
expression profiles as revealed by the microarray analysis described in this paper. The log
2
normalized expression values were used to represent
the gene specific microarray profiles.
Deveshwar et al. BMC Plant Biology 2011, 11:78

/>Page 11 of 20
wheat, and barley genomes [12]. The expression of the
rice homologues, identified by tBLASTx, showed that
several of these have specific expre ssion/significant up-
regulation in anthers, though the majority were also
expressed i n other tissues/st ages of deve lopment (Figure 7).
Four of the five annotated cohesin genes showed similar
expression levels in most tissues, but among these SMC3
and SCC3 had reduced expression in roots and TPA. The
double stran d break linked genes SPO11-1 and SPO11-2,
though expressed at relatively lower levels, showed
(a) (b)
4500
4000
3500
3000
2500
2000
1500
1000
500
0
500
1000
1500
2000
2500
3000
319
1099

1461
2771
305
663
1915
4480
448
191
133
33
2 fold down-regulated
2 fold up-regulated
specifically triggered
at the stage
PMA vs SAM
SCP vs MA
TPA vs SCP
MA vs PMA
Number of Genes
Comparison of adjacent stages
Amino acid Metabolism
Car bohydrate Me tabolism
CatalyƟc AcƟvity
Cell Cycle
Cell Structure
ChromaƟn remodelling
Energy Metabolism
Protein Metabolism
Lipid Metabolism
NucleoƟde Metabolism

Pr otein - Protein InteracƟon
RNA Metabolism
Secondary Metabolism
Signal TransducƟon
Stress
TranscripƟon Factor
TranslaƟon
Transporter
Vesicular Trafficking
0
10
18
16
14
12
10
8
6
4
2
0
12
10
8
6
4
2
0
12
10

8
6
4
2
9
8
7
0
6
5
4
3
2
1
PMA Activate
d
PMA Up
MA Activated
MA Up
SCP Activated
SCP Up
TPA Activated
TPA Up
Percentage of genes
Figure 6 Analysis of gene activation during anther development. The transcriptomes of all four anther stages were compared to their
preceding stage of development. SAM has been used as the reference for PMA. (a) The number of genes up- or down-regulated ≥ 2-fold at p-
value ≤ 0.005 are plotted on the graph. Amongst the up-regulated genes, the numbers that have no detectable expression (GC-RMA value ≤ 10)
in any of their previous anther stages as well as non-anther stages have been annotated in maroon boxes in the individual columns as
specifically ‘triggered’. While such candidates may have expression in later anther stages, expression first appears in that particular anther stage.
(b) A bar graph highlighting the distribution of the up-regulated and specifically-triggered genes at individual stages of development into

functional categories based on GO annotations.
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 12 of 20
specifically higher transcript accumulation in PMA, MA
and TPA. S ome components of the mismatch repair
machinery (MSH2, MSH6/7), RAD52 epistasis group
(AtRAD51,ScRAD51, RAD51B, RAD50)andthose
involved in recombination/synapsis (SC P1, MND1,
DMC1, ZYP1A/1B, MUS81) also exhibited higher tran-
script accumulation in stages of anther development. The
yeast meiosis-related genes described above were also
analyzed for their expression profiles during vegetative,
pre-meiotic and stages after induction of meiosis in yeast
(Additional File 7), utilizing the microarray data (GEO
Accession no. GSE18256) as described [30]. Of 21 g enes
analyzed, 11 showed uniform expression in vegetative as
well as sporulation stages, while the remaining 10 genes
were up regulated by at least two folds during meiosis in
yeast (written in red in Additional File 7). Most of the
rice orthologu es of meiosis-related yeast genes were also
found to show enhanced transcript accumulation in
anther stages.
Functions for only a few me iosis-related genes,
selected on the basis of homology to known genes in
other systems or meiosis-affecting mutant phenotypes,
have been validated in rice. These genes, with the excep-
tion of AtRCK, exhibit a characteristic meiotic anther
specific expression profile (Figure 7, lowermost panel).
Our co-expression analysis revealed two cluster (7 and
8; containing 1105 genes) exhibiting meiotic anther pre-

ferential expression profiles(Figure2).Includedinthis
list are ZEP1 [31], DMC1 [32]and MEL2 [33] which
have been implicated in transition from mitotic to meio-
tic cell division and synapsis of homologous chromo-
somes. The genes in this cluster, therefore, could be a
valuable resource for mining other components of meio-
tic machinery and meiosis related regulatory networks.
Proportion of putative sperm cell expressed genes in the
TPA transcriptome
With the aim of identi fying genes contributing to sperm
cell transcriptome in rice, we performed a comparative
analysis of Arabidopsis, maize and lily sperm cell/gen-
erative cell expressed genes with the TPA transcriptome
[34-36]. We then complementedthisanalysisbyover-
laying information on genes with specific expression in
rice anthers in order to determine sperm cell-specific
genes from the TPA transcriptome. BLASTx analysis
identified rice homologues for 338 genes from the maize
sperm cell, 3,152 from the Arabidopsis sperm cell, and
241 from the lily generative cell transcriptome that were
represented on the Rice Genome Array (Additional File
2). Hobo et al., [16] identified 28,141 anther-ex pressed
genes in rice and classified them into 20 clusters based
on co-expression profiles; five of which included genes
with expression in rice bi-cellular and tri-cellular micro-
spores. The 5,345 genes in these five clusters were also
included in this analysis (Additional File 2). Rice homo-
logues of 90.5% maize, 86.7% lily and 82.2% Arabidopsis
germline-expressed genes were represented in the rice
TPA transcriptome. Even the sperm-cell trans criptome

datasets obtained from maize and Arabidopsis that
represen t two evolutionary diverse plan t grou ps (mono-
cots and dicots), had proportional representation in the
rice TPA transcriptome (Figure 8). However, when the
maize and Arabidopsis sperm cell transcriptomes were
2.5 7.5 12.5
ZIP1
MEC1
RFA2
MND1
SCP1
MEK1
RFA1
AHP2
DMC1
ZYP1A/ZYP1B
ASY1
AtDMC1
MUS81
MER3
SAE2
MSH4
Recombination/Synapsis/Pairing
RAD54
RAD59
MRE11
RAD51 (At)
RAD51 (Sc)
RAD51B
RAD50

RAD51C
RAD51D
RAD 52 Epistasis Group
MLH1
MSH3
MLH2
MSH2
MSH6/7
Mismatch Repair Machinery
SPO11-2
SPO11-1
Double Strand Break Formation
SMC3
SCC3
SCC1
DIF1/SYN1
AtSDS
AtRCK
SMC1
Cohesins
R
YL
ML
Sdl
SAM
PMA
MA
SCP
TPA
S1

S2
S3
S4
S5
LOC_Os03g50220
LOC_Os09g32930 (PAIR2)
LOC_Os03g58600(OsMEL1)
LOC_Os05g50410 (OsRAD21-4)
LOC_Os03g12414(OsSDS)
LOC_Os02g40450(OsRCK)
LOC_Os11g04954
LOC_Os01g42880
LOC_Os06g03682
LOC_Os02g40450
LOC_Os01g72880
LOC_Os02g37920
LOC_Os09g10850
LOC_Os04g54340
LOC_Os05g19270
LOC_Os01g08540
LOC_Os01g71960
LOC_Os02g29464
LOC_Os11g40150
LOC_Os12g31370
LOC_Os05g03050
LOC_Os01g39630
LOC_Os09g01680
LOC_Os12g40890
LOC_Os11g19250
LOC_Os02g53680

LOC_Os02g42230
LOC_Os06g41050
LOC_Os07g31300
LOC_Os05g09620
LOC_Os07g12910
LOC_Os12g44390
LOC_Os02g04040
LOC_Os03g54091
LOC_Os08g06050
LOC_Os02g48010
LOC_Os04g37960
LOC_Os04g58630
LOC_Os07g30240
LOC_Os12g04980 (OsDMC1)
LOC_Os03g01590 (PAIR1)
LOC_Os10g26560 (PAIR3)
Characterized Genes in Rice
Figure 7 Expression profiles of putative homologues of known
meiosis related genes in yeast and/or Arabidopsis that were
identified by sequence similarity searches (see reference [20])
during various developmental stages in rice. Gene names as
taken from the respective sources are shown on the left, while
locus IDs of putative homologues in rice are given on the right side
of each expression profile. The lowermost panel shows the
expression profiles of genes whose functional association in rice has
been validated.
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 13 of 20
compared with each o ther, only 151 genes were found
to be common, amounting to 44.7% and 3.6% of their

respective transcriptomes (Additional File 2). In all,
3,662 rice homologues of sperm cell expressed genes in
other systems were identified which would comprise the
putative sperm cell tra nscriptome of rice (Figure 8, indi-
cated by the total number of transcripts delineated
within the red dashed line).
Discussion
The microarray data presented here forms a robust plat-
form for the studies on developmental and m olecular
aspects of male gametophyte development in rice and in
cereals at large. A high degree of correlation obtained
between the three biological replicates for all stages
investigated underlines the reproducibility and strength
of the data, which has also been validated by Q-PCR
and in-situ hybridization analyses. The MAS 5.0 based
present calls representing the size of the transcriptomes
(15,465 - 18,090) examined, was found to be signifi-
cantly higher in comparison to a recent sequencing-by-
synthesis based analysis of rice anther transcriptomes in
which about 3000 - 12000 distinct transcripts were
detected in individual stages of anther development [37].
On the other hand it was much less when compared to
another recent study of the rice anther transcriptome
using the same platform (Affymetrix) [38] in whic h
30,186 - 28,280 probe-sets were reported. The difference
could have resulted from the fact that we ha ve used a
more refined sub-set of genes where the redundant
probe-set IDs and genes coding for transposable ele-
ments (TEs) were removed. The size of maize anther
transcripto mes (based on 44 K ma ize array), however,

was found to have a comparable number of transcripts
to those found in our analysis [39]. Compariso n of the
anther transcriptomes revealed a high correlation
between SAM, PMA, MA and SCP stages, indicating
that there are subtle differences in expression across
TPA transcriptome (15,465)
12
22
660
1001
(1)
5
6
23
(1)
59
67
30
(1)
2,214
(4)
962
(26)
3,143
(282)
66
(2)
43
(1)
1

1
55
2
11
14
81
7
6
1
4
87
(1)
8,660
(133)
Rice homologues of Lilly Generative
cell transcriptome, 241
[36]
Rice homologues of maize
sperm cell library, 338
[35]
Rice homologues of Arabidopsis
sperm cell library, 4,152
[34]
Rice genes from microspore
expressed clusters, 5,345
[16]
Figure 8 Identification of putative male gamete transcripts in rice. The Venn diagram shows overlap between genes that were identified as
being present in TPA with microspore preferential genes [16], and homologues found by sequence similarity in the Arabidopsis sperm cell
transcriptome [34], maize sperm cell ESTs [35] and the lily generative cell transcriptome [36]. The number of genes from the respective
transcriptomes that could be mapped on the Rice Genome Array are bold and in italics, while the number of genes that are specifically

expressed in the rice TPA transcriptome are indicated in parentheses. The red dashed line constitutes the total number of rice homologues
(excluding those in parentheses) that contribute to the putative sperm cell transcriptome in rice which have been identified from the other
systems examined.
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 14 of 20
these three stages. The high similarity observed between
SAM a nd PMA suggests that changes in the expression
of very few genes are required to trigger anther develop-
ment in rice. Most of the differentially expressed genes
identified were found to be regulatory in nature, there-
fore, although these changes are few i n number they
can potentially initiate a chain of events in subsequent
stages to influence expression of m any downstream
genes. It would seem that PMA is therefore a transitory
phase, where the decision to undergo meiosis is taken in
some specialized cells. PMA and MA represent very
early stages of anther development in which there is
relatively high representation of sporophytic tissue in
comparison to the gametophytic tissue, with most of the
transcriptome changes corresponding to sporogenous
tissue and the developing tapetum. TPA, however, con-
tains a relatively higher cellular mass that represents
mature gametophytic tissue. In the current study, we
clearly show marked differences in its transcript consti-
tution from the rest of the anther transcript omes inves-
tigated. The TPA stage is also characterized by the
smallest and the most diverse transcriptome of the four
stages analyzed. This could be by virtue of the distinc-
tive transcriptomes of the male gametophyte and sperm
cells [34] and down-regulation of a large number of

genes that might not be required for the development of
the gametophyte [39].
Comparison of our data with the recently published
transcriptome of haploid male g ametophyte develop-
ment [40] substantiates our staging of anther develop-
ment. Of the 188 unique probe-sets enriched in
tricellular pollen (TCP) as identified by Wei and cowor-
kers [40] 160 (85%) were expressed in TPA. Also of the
525 uni-nucleate microspore (UNM) enriched probe-
sets, 405 (77%) were expressed in SCP.
In the present study, we have grouped differentially
expressed genes into 10 co-expression groups. The
information gathered from frequently co-expressed
genes across multiple datasets and across different
organisms has previously been used to verify gene inter-
action patterns and also to predict novel gene interac-
tion networks [41-44]. A large number of genes ( ~34%)
found in these co-expression groups have not even been
annotated and even fewer of those that are annotated
have been validated for their role in a nther develop-
ment. Therefore, identification of genes in these c o-
expression clusters pave the way for more focused inves-
tigations leading to a better understanding of gene regu-
latory networks.
The list of specifically activated transcription factors in
PMA included four trans cription factors (DRD1, ZOS2-
03-C
2
H
2

zinc finger, a helix-loop-helix, and a MYB tran-
scription factor). Also, eleven genes (OsMADS1, 2, 3, 4,
5, 6, 7, 8, 17, 34 and 58) belonging to the MADS-box
family were up regulated by 2 folds in PMA with respect
to SAM, which were also shown to be part of the pollen
mother cell preferential transcriptome ([45], A dditio nal
File 6). Some of these MADS box genes have also been
implicated in anther development [46,47]. Four mem-
bers of the YABBY gene family were found to be specifi-
cally down regulated in PMA in comparison to SAM.
However, in the MA, the NAM and AP2 class of genes
dominated the list of differentially expressed transcrip-
tion factors. While eight NAM and three AP2 family
genes were up regulated in MA, eight AP2 genes were
down regulated (Additional File 6). In SCP, besides a
shuffling in the pool of transcription factors, down-regu-
lation of the translation machinery was observed with
more than 200 translation-related genes significantly
affected by more than two folds. Most of these genes
continued with the downward trend in TPA as well. In
concurrence with previous obse rvations from several
groups [13,48,49] most of the down-regulated genes
code for ribosomal proteins and elongation f actors.
There were also genes coding for LSM (like-Sm) domain
containing and RNA recognition motif proteins that are
known for their involvement in p re-mRNA processing
[50,51].
A major proportion of the differentially expressed sig-
nal transduction component s included those involved in
calcium-mediat ed signaling, e.g. calcium dependent pro-

tein kinases, caleosins and other protei ns containing C2
domain and EF hands. Genes involved in secondary
metabolism (for example, those coding cytochrome 450,
chalcone flavonone isomerase, strictosidinesynthase)
showed marked up-regulationinMAandSCP.During
meiosis and the single-celled microspore stage, tapetal
cells are most active and are known to be involved in
the synthesis of flavonoids and other secondary metabo-
lites that eventu ally find their way to developing micro-
spores [52], therefore, up-reg ulation of secondary
metabolism related genes could in fact be related to
tapetum development. Though not significant in num-
ber, genes involved in chromatin remodeling were also
differentially expressed during progression of anther
development. In addition, genes expressed in response
to various abiotic stresses e.g., those coding for late
embryogenesis abundant (LEA) proteins, dehydrins, and
other s enescence-associated proteins showed stage-spe-
cific differential expressio n, emphasizing that there may
be parallels between the molecular mechanisms involved
in reproductive development and stress (deta ils of these
genes can be found in Additional File 6).
Another interesting observation was that the anther
transcriptome showed a high level of similarit y with the
seed transcriptome profile, when compared with other
stages/tissues. The similarities in the two organs could
be due to biochemical processes that are common to
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 15 of 20
them. Both anthers and seeds a re metabolically active

tissues that exhibit high rates of cell division and both
act as sinks for sugar derivatives, which are converted to
starch at a rapid rate in these tissues. Furthermore, we
have recently analyzed t he similarities between repro-
ductive developmental sta ges and de hydration stress in
rice [53], where the findings indicate a high degree of
overlap between genes that show differential expression
during mature stages of panicle development, natural
desiccation of seeds and plants expo sed to dehydration
stress. The findings in the present analysis highligh t that
both anthers and seeds could be utilizing similar regula-
tory networks for accumulation of starch, as th ey enter
a phase of biological desiccation.
Gene regulation by means of RNA interference has been
shown to play a vital role in anther d evelopment [54].
Reportshavealsoshownthepresenceoffunctional
miRNA i n late stages of a nther development [55]. Our data
has also revealed up-regula tion of genes coding for argo-
nautes and other proteins with PAZ and PIWI domains in
pre-meiotic and meiotic anthers in a stage preferential
manner, suggesting that a different subset of RNAi
machinery might be activated in reproductive tissues espe-
cially during early a nther development. A rice pre-meiosis-
specific argonaute gene OsMEL1 (LOC_Os03g58600) that
has recently been implicated in male meiosis, [24] was
shortlisted in our analysis and identified as being PMA-
specific. It might therefore be interesting to explore the
function of other similar components, determining
whether there is reproduction-specific RNAi machinery.
Towards understanding expression dynamics of the

meiome
Meiosis has long been the subject of research with pio-
neeri ng investigation in yeast (Saccharomyces cerevis iae)
[56]. In flowering plants, identification of many meiotic
mutants in Arabidopsis, rice and maize has led to func-
tional characterization of close to fifty plant meiotic
genes, that has substantially added to our current under-
standing of genes involved in plant meiosis [57,58]. The
meiosis related genes characterized in rice include
PAIR1 (HOMOLOGOUS PAIRING ABERRATION IN
RICE MEIOSIS1)[59],PAIR2 [homologous to Saccharo-
myces cerevisiae HOP1 (HOMOLOGOUS PAIRING 1)
and Arabidopsis ASY1 (ASYNAPTIC1)] [60], OsRAD21-4
(RADIATION SENSITIVE 21-4)[27],OsDMC1 (DIS-
RUPTION OF MEIOTIC CONTROL 1) [61], PAIR3 [62],
OsMEL1(MEIOSIS ARRESTED AT LEPTOTENE1)[24],
OsSDS (SOLO DANCERS) and OsRCK (ROCK-N-ROLL-
ERS) [63]. All these genes except PAIR1, PAIR3 and
OsMEL1 were identified due to their homology to meio-
sis-related genes in yeast or Arabidopsis. However unlike
in yeast and mammalian systems, we are still far from
constituting the plant meiome.
Our data shows that a large majority of rice meiosis
homologues (from yeast and mammalian systems) do
not express in a meiosis-specific manner. For example,
AtAHP2 [64] is known to be involved in meiosis but it
is expressed in other vegetative tissues as well. Likewise
AtRAD51 is expressed in other stages but it has been
shown to be essential for the progression of normal
meiosis [65]. It could either mean that a greater pro por-

tion of genes involved in meiosis play a role in other
cellular functions as well or that other genes may have
taken up meiosis-specific functions in plan ts. Some
meiotic gene s have been shown to be plant -specific (for
example, Poor Homologous Synapsis ; PHS1 [66]).
Recently, Tang and coworkers [ 45] carried out global
expression profiling of laser-captured pollen mother
cells (PMCs) in rice using the 44k Agilent array. By
comparing the e xpression of PMC expressed genes to
those expressed in seedlings and tricellular pollen they
could identify 1,158 PMC-preferential genes. These
PMC preferentially expressed genes contained many
known meiotic genes, including OsSPO11-1 [67],PAIR1
[59],PAIR2[60],PAIR3[62],OsDMC1[61],OsMEL1
[24], OsRAD21-4 [27], OsSDS [63], and ZEP1 [31]. Since
917 of the 1158 PMC-preferential genes were repre-
sented in our da ta set (which is based on the Affymetrix
57k chip) we decided to analyze the expression profiles
of these g enes in all four stages of a nthers (see Figure
9a; Additional File 8). Interestingly, of the 917 genes,
702 expressed bo th in pre-meiotic (PMA) and meiotic
ant hers (MA) (Figure 9b) and 561 of these expres sed in
SAM as well, albeit at relatively lower levels. However,
when this data set of 917 genes was pa rsed through our
data set of anther-specific genes, we could identify only
67 genes that were expressed in PMA and MA (44 were
expressed in both PMA and MA, two were PMA-speci-
fic and 21 were expressed specifically in MA). Further-
more, most of the 702 genes that were expressed in
both PMA and MA were also expressed a t significant

levels in other stages of anther development (Figure 9b).
These observations strengthen our hypothesis that the
expression for the majority of meiosis-related genes is
not restricted t o cells undergoing meiosis and that they
may
participate in functions other than meiosis or that
other genes may have taken up meiosis-specific func-
tions in plants. Therefore, the 372 genes constituting
the meiotic anther specific expression profile (Figure 2)
in this study should serve as a valuable resource for
mining, the as yet unidentified, compo nents of the
meiotic machinery and associated regulatory networks
in rice.
Differential expression analysis involving comparison
of adjacent stages revealed an interesting pattern show-
ing a steady and significant increase in the number of
genes activating in post-meiotic stages (i.e., in SCP and
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 16 of 20
TPA). In these stages a total of 4,232 transcripts were
up regulated in comparison to 1, 418 in PMA and MA
combined. Additionally, the post-meiotic stages of SCP
and TPA contained a significantly larger proportion of
genes down regulated (6,395) when compared to the
earlier stages of PMA and MA (968). A similar trend of
a large n umber of ge nes getting transcriptionally acti-
vated and deactivated in post-meiotic anthers was also
observed in maize [39]. In maize, 867 genes were up
regulated while 908 were down regulated in post-meio-
tic stages of anther development. We detected ortholo-

gues for 345 and 346 genes, respectively, in our data set,
and of these 265 (~77%) and 226 (~65%) exhibited simi-
lar expression profiles in rice. Incidentally, a large per-
centage of the down-regulated genes in both rice and
maize anthers are those t hat are expressed at significant
levels (normalized average expression value ≥ 50) (data
not shown), which suggests that meiosis may act as a
two-way molecular switch that activates a large number
of gametophytic genes, and at the same time, shuts
down the sporophytic machinery that is presumably not
necessary for male gametophyte development.
Sperm cell transcriptome
The transcriptomes of pollen and sperm cells have be en
reported to be smaller than and distinct from those of
vegetative tissues [34,35]. By comparing known sperm
cell transcriptomes with the TPA transcriptome, we
have attempted to identify the constitution of the rice
sperm cell transcript pool. We show that a large propor-
tion of transcripts constituting the Arabidopsis sperm
cell (82.3%), maize sperm cell (90.1%) and the lilly gen-
erative cell (86.7%) transcriptomes were represented in
the rice TPA transcriptome, suggesting a high degree of
similarity between sperm cell transcriptomes of mono-
cot and dicot plants. Categorization of TPA activated
and up-regu lated genes into GO functional groups (Fig-
ure 6) indicates that genes encoding signal transduction
components, cell structure components, transporters,
transcription factors and stress related pathways, could
be the major contributors to the sperm cell transcrip-
tome. Since 448 of the TPA expressed genes have not

been previously reported to be expressed in v egetative
tissues and throughout the s tages of seed development,
they could serve as a useful resource to mine putative
sperm cell transcripts for validation of their function in
this unique cell type.
Conclusions
Implications in defining components of biochemical and
gene regulatory networks
Identification of co-expressing clusters in a developmen-
tal event is indicative of common or related regulatory
pathways. Co-expression is often related to co-regula-
tion, and genes that follow similar expressio n profiles
may be the targets of the same transcription factors.
SAM
PMA
MA
SCP
TPA
(a)
(b)
2.03 8.71 15.39
PMA
(739)
PMC
Preferential
(917)
29
(2)
49
(21)

141
(37)
8
(0)
4
(0)
561
(7)
12
(0)
MA
(755)
SAM
(585)
Figure 9 Comparison of the pollen mot her cell transcriptome with anther stages. (a) Venn diagram showing the expression of pollen
mother cell (PMC) preferential genes identified by Tang and co-workers [45] in PMA, MA and SAM. The number of probe-sets expressed in each
stage is indicated, with the number of genes specifically expressed in anthers indicated in parentheses. (b) A heat-map representing the
expression profiles of 702 PMC preferential genes (from the original 917 identified in comparing the 44K and 57K chip - see reference [45]and the
discussion) that are expressed in SAM and the four stages of anther development (PMA, MA, SCP and TPA).
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 17 of 20
Our studies have allowed identification of specifically
regulated genes. Comparison with four of the vegetative
stages and five seed development stages of ri ce has
allowed segregation of tran scripts dedicated to anther
development function, especially the three co-expression
groups that identify genes showing expression peaks in
the four stages of anthers (Figure 2). Many of the genes
in these c lusters have not yet been annotated and as
such , deserve further attention as a source of genes that

could have (as yet) unidentified roles in meiosis and
other stages examined. Such analy sis is of great signifi-
cance for future research, with several candidates now
being targeted for studies that will build towards our
understanding of regulatory networks and validating
gene function(s).
Additional material
Additional File 1: Box-Whisker plot showing the range of
expression of Magnoporthe genes across the 14 stages of
development used in the microarray. LM, Mature Leaf; LY, Y-Leaf; R,
Root; SDL, 7 day old seedling; SAM, Shoot apical meristem; PMA, Pre-
meiotic anther; MA, Meiotic anther; SCP, Anthers with single celled
pollen; TPA, Anther with trinucleate pollen; S1-S5, Seed stages from 0
days after pollination (DAP) till 30 DAP.
Additional File 2: List of rice homologues for Arabidopsis, lily and
maize sperm cell transcripts and significance values from BLASTx
searches and their comparison with the TPA transcriptome.
Additional File 3: List of Primers used in Real-time PCR.
Additional File 4: Scatter plots comparing gene expression of four
stages of anther development as well as shoot apical meristem
(SAM). Numerical figures in the blocks show the number of genes with
at least 2-fold differential expression between the stages. The correlation
co-efficient for gene expression between the stages is indicated at the
top of each plot. Clearly, PMA (pre-meiotic anther), MA (meiotic anther)
and SCP (single-celled pollen) have more similarity in their transcriptome
than TPA (tri-nucleate pollen), which shows higher variation in
transcripts.
Additional File 5: MAS5 detection calls and p-values for the list of
unique probe-set IDs and probe-set lists of SAM, PMA, MA, SCP,
seed, and leaf transcriptomes.

Additional File 6: Raw and log
2
transformed expression values,
probe set IDs, Locus IDs, functional categories, putative functions
and cluster categorization of 11,915 genes differentially expressed
in anthers.
Additional File 7: Expression profiles of early meiosis genes in
yeast. Numbers on the Y axis are normalized. The data was normalized
for the minimum value as zero. Meiotic time points are shown on the X
axis. Gene names written in red are at least 1.95-fold up regulated when
comparing the maximum value of meiotic verses maximum value of the
non-meiotic stages. Fold changes are shown in parentheses. The
expression data was obtained from GEO accession number GSE18181.
Additional File 8: A comparison of the pollen mother cell
transcriptome with that of SAM, PMA, MA, SCP and TPA.
Acknowledgements
We thank Prof. Akhilesh K. Tyagi and Dr. Meenu Kapoor for stimulating
discussions, and Drs. Ramesh Hariharan and Sanjeev Singh for their help in
microarray data analysis. We gratefully acknowledge the financial support
provided by the Department of Science and Technology (DST), Govt. of
India and the Australian Government under the Australia-India Strategic
Research Fund - a component of the Australian Scholarships initiative. PD and
RS acknowledge the Council of Scientific and Industrial Research (CSIR) for
the Senior Research fellowships. We would also like to thank the anonymous
referees for their helpful comments through the review process, which
enabled the paper to be improved.
Microarray Data Submission
The microarray data of the four anther stages has been submitted to the
Gene Expression Omnibus (GEO; under
the GSE27726 accession series. Accession numbers for the *.cel files of other

stages/tissues of rice development used in this paper are GSE6893 and
GSE6901.
Author details
1
Interdisciplinary Centre for Plant Genomics and Department of Plant
Molecular Biology, University of Delhi, South Campus, New Delhi - 110021,
India.
2
Waite Research Institute and School of Agriculture, Food & Wine, The
University of Adelaide, Waite Campus, PMB1, Glen Osmond, South Australia
5064, Australia.
3
Current address: Department of Plant Pathology, University
of California, Davis, CA, USA.
Authors’ contributions
PD carried out the microarray experiments for PMA, SCP and TPA stages,
performed in situ hybridization experiments, analyzed the data and drafted
the manuscript. RS identified the anther stages and performed the
microarray experiment for MA stage. WDB and JAA helped in the data
analysis and preparation of the manuscript. SK conceptualized and designed
the experiments, contributed to the data analysis and manuscript writing. All
authors read and approved the final manuscript.
Received: 15 December 2010 Accepted: 9 May 2011
Published: 9 May 2011
References
1. Goldberg RB, Beals TP, Sanders PM: Anther development: basic principles
and practical applications. Plant Cell 1993, 5:1217-1229.
2. Yang SL, Xie LF, Mao HZ, Puah CS, Yang WC, Jiang L, Sundaresan V, Ye D:
Tapetum determinant1 is required for cell specialization in the
Arabidopsis anther. Plant Cell 2003, 15:2792-2804.

3. Ma H: Molecular genetic analyses of microsporogenesis and
microgametogenesis in flowering plants. Annu Rev Plant Biol 2005,
56:393-434.
4. Yu-Jin Sun CLH, Chen Chang-Bin, Hong Ma: Regulation of Arabidopsis
early anther development by putative cell-cell signalling molecules and
transcriptionl regulators. Journal of Integrative Plant Biology 2007, 49:60-68.
5. Borg M, Brownfield L, Twell D: Male gametophyte development: a
molecular perspective. J Exp Bot 2009, 60:1465-1478.
6. Wilson ZA, Zhang DB: From Arabidopsis to rice: pathways in pollen
development. J Exp Bot 2009, 60:1479-1492.
7. Zhao X, de Palma J, Oane R, Gamuyao R, Luo M, Chaudhury A, Herve P,
Xue Q, Bennett J: OsTDL1A binds to the LRR domain of rice receptor
kinase MSP1, and is required to limit sporocyte numbers. Plant J 2008,
54:375-387.
8. Zhao DZ, Wang GF, Speal B, Ma H: The excess microsporocytes1 gene
encodes a putative leucine-rich repeat receptor protein kinase that
controls somatic and reproductive cell fates in the Arabidopsis anther.
Genes Dev 2002, 16:2021-2031.
9. Nonomura K, Miyoshi K, Eiguchi M, Suzuki T, Miyao A, Hirochika H, Kurata N:
The MSP1 gene is necessary to restrict the number of cells entering into
male and female sporogenesis and to initiate anther wall formation in
rice. Plant Cell 2003, 15:1728-1739.
10. Canales C, Bhatt AM, Scott R, Dickinson H: EXS, a putative LRR receptor
kinase, regulates male germline cell number and tapetal identity and
promotes seed development in Arabidopsis. Curr Biol 2002, 12:1718-1727.
11. Kinoshita K, Obayashi T: Multi-dimensional correlations for gene
coexpression and application to the large-scale data of Arabidopsis.
Bioinformatics 2009, 25:2677-2684.
12. Lee TH, Kim YK, Pham TT, Song SI, Kim JK, Kang KY, An G, Jung KH,
Galbraith DW, Kim M, Yoon UH, Nahm BH: RiceArrayNet: a database for

correlating gene expression from transcriptome profiling, and its
application to the analysis of coexpressed genes in rice. Plant Physiol
2009,
151:16-33.
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 18 of 20
13. Honys D, Twell D: Transcriptome analysis of haploid male gametophyte
development in Arabidopsis. Genome Biol 2004, 5:R85.
14. Crismani W, Baumann U, Sut ton T, Shirley N, W ebster T, Spange nber g G,
Langridge P, Able JA: Microarr ay expression analysis of meiosis and
microsporogenesis in hexaploid bread wheat. BMC Genomics 2006,
7:267.
15. Wang Z, Liang Y, Li C, Xu Y, Lan L, Zhao D, Chen C, Xu Z, Xue Y, Chong K:
Microarray analysis of gene expression involved in anther development
in rice (Oryza sativa L.). Plant Mol Biol 2005, 58:721-737.
16. Hobo T, Suwabe K, Aya K, Suzuki G, Yano K, Ishimizu T, Fujita M, Kikuchi S,
Hamada K, Miyano M, Fujioka T, Kaneko F, Kazama T, Mizuta Y, Takahashi H,
Shiono K, Nakazono M, Tsutsumi N, Nagamura Y, Kurata N, Watanabe M,
Matsuoka M: Various spatiotemporal expression profiles of anther-
expressed genes in rice. Plant Cell Physiol 2008, 49:1417-1428.
17. Suwabe K, Suzuki G, Takahashi H, Shiono K, Endo M, Yano K, Fujita M,
Masuko H, Saito H, Fujioka T, Kaneko F, Kazama T, Mizuta Y, Kawagishi-
Kobayashi M, Tsutsumi N, Kurata N, Nakazono M, Watanabe M: Separated
transcriptomes of male gametophyte and tapetum in rice: validity of a
laser microdissection (LM) microarray. Plant Cell Physiol 2008,
49:1407-1416.
18. Raghvan V: Anther and Pollen Development in Rice (Oryza sativa).
American Journal of Botany 1988, 75:183-196.
19. Benjamini Y, Hochberg Y: Controlling the false discovery rate: a practical
and powerful approach to multiple testing. J R Stat Soc B 1995,

57:289-300.
20. Bovill WD, Deveshwar P, Kapoor S, Able JA: Whole genome approaches to
identify early meiotic gene candidates in cereals. Funct Integr Genomics
2009, 9:219-229.
21. Arora R, Agarwal P, Ray S, Singh AK, Singh VP, Tyagi AK, Kapoor S: MADS-
box gene family in rice: genome-wide identification, organization and
expression profiling during reproductive development and stress. BMC
Genomics 2007, 8:242.
22. Jackson D: In-situ hybridisation in plants. In Molecular Plant Pathology: A
Practical Approach Edited by: Bowles DJ, Gurr SJ, McPherson M: Oxford
University Press 1991, 163-174.
23. Agarwal P, Arora R, Ray S, Singh AK, Singh VP, Takatsuji H, Kapoor S,
Tyagi AK: Genome-wide identification of C2H2 zinc-finger gene family in
rice and their phylogeny and expression analysis. Plant Mol Biol 2007,
65:467-485.
24. Nonomura K, Morohoshi A, Nakano M, Eiguchi M, Miyao A, Hirochika H,
Kurata N: A germ cell specific gene of the ARGONAUTE family is
essential for the progression of premeiotic mitosis and meiosis during
sporogenesis in rice. Plant Cell 2007, 19:2583-2594.
25. Luo H, Lee JY, Hu Q, Nelson-Vasilchik K, Eitas TK, Lickwar C, Kausch AP,
Chandlee JM, Hodges TK:
RTS,
a rice anther-specific gene is required for
male fertility and its promoter sequence directs tissue-specific gene
expression in different plant species. Plant Mol Biol 2006, 62:397-408.
26. Zhang D, Liang W, Yin C, Zong J, Gu F: OsC6, encoding a lipid transfer
protein, is required for postmeiotic anther development in rice. Plant
Physiol 2010, 154:149-162.
27. Zhang L, Tao J, Wang S, Chong K, Wang T: The rice OsRad21-4, an
orthologue of yeast Rec8 protein, is required for efficient meiosis. Plant

Mol Biol 2006, 60:533-554.
28. Nonomura KI, Nakano M, Murata K, Miyoshi K, Eiguchi M, Miyao A,
Hirochika H, Kurata N: An insertional mutation in the rice PAIR2 gene, the
ortholog of Arabidopsis ASY1, results in a defect in homologous
chromosome pairing during meiosis. Mol Genet Genomics 2004,
271:121-129.
29. Li N, Zhang DS, Liu HS, Yin CS, Li XX, Liang WQ, Yuan Z, Xu B, Chu HW,
Wang J, Wen TQ, Huang H, Luo D, Ma H, Zhang DB: The rice tapetum
degeneration retardation gene is required for tapetum degradation and
anther development. Plant Cell 2006, 18:2999-3014.
30. Berchowitz LE, Hanlon SE, Lieb JD, Copenhaver GP: A positive but complex
association between meiotic double-strand break hotspots and open
chromatin in Saccharomyces cerevisiae. Genome Res 2009, 19:2245-2257.
31. Wang M, Wang K, Tang D, Wei C, Li M, Shen Y, Chi Z, Gu M, Cheng Z: The
central element protein ZEP1 of the synaptonemal complex regulates
the number of crossovers during meiosis in rice. Plant Cell 2010,
22:417-430.
32. Kant CR, Rao BJ, Sainis JK: DNA binding and pairing activity of OsDmc1,a
recombinase from rice. Plant Mol Biol 2005, 57:1-11.
33. Nonomura K, Eiguchi M, Nakano M, Takashima K, Komeda N, Fukuchi S,
Miyazaki S, Miyao A, Hirochika H, Kurata N: A novel RNA-recognition-motif
protein is required for premeiotic G1/S-phase transition in rice (Oryza
sativa L.). PLoS Genet 2011, 7:e1001265.
34. Borges F, Gomes G, Gardner R, Moreno N, McCormick S, Feijo JA, Becker JD:
Comparative transcriptomics of Arabidopsis sperm cells. Plant Physiol
2008, 148:1168-1181.
35. Engel ML, Chaboud A, Dumas C, McCormick S: Sperm cells of Zea mays
have a complex complement of mRNAs. Plant J 2003, 34:697-707.
36. Okada T, Singh MB, Bhalla PL: Transcriptome
profiling of Lilium

longiflorum generative cells by cDNA microarray. Plant Cell Rep 2007,
26:1045-1052.
37. Huang MD, Wei FJ, Wu CC, Hsing YI, Huang AH: Analyses of advanced rice
anther transcriptomes reveal global tapetum secretory functions and
potential proteins for lipid exine formation. Plant Physiol 2009,
149:694-707.
38. Fujita M, Horiuchi Y, Ueda Y, Mizuta Y, Kubo T, Yano K, Yamaki S, Tsuda K,
Nagata T, Niihama M, Kato H, Kikuchi S, Hamada K, Mochizuki T, Ishimizu T,
Iwai H, Tsutsumi N, Kurata N: Rice expression atlas in reproductive
development. Plant Cell Physiol 2010, 51:2060-2081.
39. Ma J, Skibbe DS, Fernandes J, Walbot V: Male reproductive development:
gene expression profiling of maize anther and pollen ontogeny. Genome
Biol 2008, 9:R181.
40. Wei LQ, Xu WY, Deng ZY, Su Z, Xue Y, Wang T: Genome-scale analysis and
comparison of gene expression profiles in developing and germinated
pollen in Oryza sativa. BMC Genomics 2010, 11:338.
41. Voy BH, Scharff JA, Perkins AD, Saxton AM, Borate B, Chesler EJ,
Branstetter LK, Langston MA: Extracting Gene Networks for Low-Dose
Radiation Using Graph Theoretical Algorithms. PLoS Comput Biol 2006, 2:
e89.
42. Stuart JM, Segal E, Koller D, Kim SK: A gene-coexpression network for
global discovery of conserved genetic modules. Science 2003,
302:249-255.
43. Lee HK, Hsu AK, Sajdak J, Qin J, Pavlidis P: Coexpression analysis of human
genes across many microarray data sets. Genome Res 2004, 14:1085-1094.
44. Basso K, Margolin AA, Stolovitzky G, Klein U, Dalla-Favera R, Califano A:
Reverse engineering of regulatory networks in human B cells. Nat Genet
2005, 37:382-390.
45. Tang X, Zhang ZY, Zhang WJ, Zhao XM, Li X, Zhang D, Liu QQ, Tang WH:
Global gene profiling of laser-captured pollen mother cells indicates

molecular pathways and gene subfamilies involved in rice meiosis. Plant
Physiol 2010, 154:1855-1870.
46. Kater MM, Dreni L, Colombo L: Functional conservation of MADS-box
factors controlling floral organ identity in rice and Arabidopsis. J Exp Bot
2006, 57:3433-3444.
47. Zhang DB, Wilson ZA: Stamen specification and anther development in
rice. Chinese Science Bulletin 2009, 54:2342-2353.
48. Becker JD, Boavida LC, Carneiro J, Haury M, Feijo JA: Transcriptional
profiling of Arabidopsis tissues reveals the unique characteristics of the
pollen transcriptome.
Plant Physiol 2003, 133:713-725.
49.
Honys D, Twell D: Comparative analysis of the Arabidopsis pollen
transcriptome. Plant Physiol 2003, 132:640-652.
50. He W, Parker R: Functions of Lsm proteins in mRNA degradation and
splicing. Curr Opin Cell Biol 2000, 12:346-350.
51. Kufel J, Allmang C, Petfalski E, Beggs J, Tollervey D: Lsm Proteins are
required for normal processing and stability of ribosomal RNAs. J Biol
Chem 2003, 278:2147-2156.
52. Kapoor S, Kobayashi A, Takatsuji H: Silencing of the tapetum-specific zinc
finger gene TAZ1 causes premature degeneration of tapetum and
pollen abortion in petunia. Plant Cell 2002, 14:2353-2367.
53. Ray S, Dansana PK, Giri J, Deveshwar P, Arora R, Agarwal P, Khurana JP,
Kapoor S, Tyagi AK: Modulation of transcription factor and metabolic
pathway genes in response to water-deficit stress in rice. Funct Integr
Genomics 2011, 11:157-178.
54. Grant-Downton R, Hafidh S, Twell D, Dickinson HG: Small RNA pathways
are present and functional in the angiosperm male gametophyte. Mol
Plant 2009, 2:500-512.
55. Fujioka T, Kaneko F, Kazama T, Suwabe K, Suzuki G, Makino A, Mae T,

Endo M, Kawagishi-Kobayashi M, Watanabe M: Identification of small RNAs
in late developmental stage of rice anthers. Genes Genet Syst 2008,
83:281-284.
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 19 of 20
56. Roeder GS: Sex and the single cell: meiosis in yeast. Proc Natl Acad Sci
USA 1995, 92:10450-10456.
57. Able JA, Crismani W, Boden SA: Understanding meiosis and the
implications for crop improvement. Functional Plant Biology 2009,
36:575-588.
58. Mercier R, Grelon M: Meiosis in plants: ten years of gene discovery.
Cytogenet Genome Res 2008, 120:281-290.
59. Nonomura K, Nakano M, Fukuda T, Eiguchi M, Miyao A, Hirochika H,
Kurata N: The novel gene HOMOLOGOUS PAIRING ABERRATION IN RICE
MEIOSIS1 of rice encodes a putative coiled-coil protein required for
homologous chromosome pairing in meiosis. Plant Cell 2004,
16:1008-1020.
60. Nonomura K, Nakano M, Eiguchi M, Suzuki T, Kurata N: PAIR2 is essential
for homologous chromosome synapsis in rice meiosis I. J Cell Sci 2006,
119:217-225.
61. Deng ZY, Wang T: OsDMC1 is required for homologous pairing in Oryza
sativa. Plant Mol Biol 2007, 65:31-42.
62. Yuan W, Li X, Chang Y, Wen R, Chen G, Zhang Q, Wu C: Mutation of the
rice gene PAIR3 results in lack of bivalent formation in meiosis. Plant J
2009, 59:303-315.
63. Chang L, Ma H, Xue HW: Functional conservation of the meiotic genes
SDS and RCK in male meiosis in the monocot rice. Cell Res 2009,
19:768-782.
64. Schommer C, Beven A, Lawrenson T, Shaw P, Sablowski R: AHP2 is required
for bivalent formation and for segregation of homologous

chromosomes in Arabidopsis meiosis. Plant J 2003, 36:1-11.
65. Li W, Chen C, Markmann-Mulisch U, Timofejeva L, Schmelzer E, Ma H,
Reiss B: The Arabidopsis AtRAD51 gene is dispensable for vegetative
development but required for meiosis. Proc Natl Acad Sci USA 2004,
101:10596-10601.
66. Pawlowski WP, Golubovskaya IN, Timofejeva L, Meeley RB, Sheridan WF,
Cande WZ: Coordination of meiotic recombination, pairing, and synapsis
by PHS1. Science 2004, 303:89-92.
67. Yu H, Wang M, Tang D, Wang K, Chen F, Gong Z, Gu M, Cheng Z:
OsSPO11-1
is essential for both homologous chromosome pairing and
crossover formation in rice. Chromosoma 2010, 119:625-636.
doi:10.1186/1471-2229-11-78
Cite this article as: Deveshwar et al.: Analysis of anther transcriptomes
to identify genes contributing to meiosis and male gametophyte
development in rice. BMC Plant Biology 2011 11:78.
Submit your next manuscript to BioMed Central
and take full advantage of:
• Convenient online submission
• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution
Submit your manuscript at
www.biomedcentral.com/submit
Deveshwar et al. BMC Plant Biology 2011, 11:78
/>Page 20 of 20